• No results found

In-hospital mortality, 30-day readmission, and length of hospital stay after surgery for primary colorectal cancer: A national population-based study

N/A
N/A
Protected

Academic year: 2021

Share "In-hospital mortality, 30-day readmission, and length of hospital stay after surgery for primary colorectal cancer: A national population-based study"

Copied!
12
0
0

Loading.... (view fulltext now)

Full text

(1)

In-hospital mortality, 30-day readmission, and length of

hospital stay after surgery for primary colorectal cancer: A

national population-based study

S. Pucciarelli

a,

*

, M. Zorzi

b

, N. Gennaro

c

, G. Gagliardi

d

,

A. Restivo

e

, M. Saugo

c

, A. Barina

a

, M. Rugge

b,f

, M. Zuin

a

,

I. Maretto

a

, D. Nitti

a

aDepartment of Surgical, Oncological and Gastroenterological Sciences, University of Padova, Padova, Italy bVeneto Tumour Registry, Veneto Region, Padova, Italy

cEpidemiological Service of the Veneto Region, Padova, Italy dDepartment of Surgery, University of Illinois at Chicago, Chicago, IL, USA

e

Department of Surgical Sciences, University of Cagliari, Cagliari, Italy f

Department of Medicine DIMED Pathology and Cytopathology Unit, University of Padova, Padova, Italy

Accepted 6 March 2017 Available online

-Abstract

Introduction: The simultaneous assessment of multiple indicators for quality of care is essential for comparisons of performance between hospitals and health care systems.

The aim of this study was to assess the rates of in-hospital mortality and 30-day readmission and length of hospital stay (LOS) in pa-tients who underwent surgical procedures for colorectal cancer between 2005 and 2014 in Italy.

Methods: All patients in the National Italian Hospital Discharge Dataset who underwent a surgical procedure for colorectal cancer during the study period were included. The adjusted odd ratios for risk factors for in-hospital mortality, 30-day readmission, and LOS were calcu-lated using multilevel multivariable logistic regression.

Results: Among the 353 941 patients, rates of in-hospital mortality and 30-day readmission were 2.5% and 6%, respectively, and the median LOS was 13 days. High comorbidity, emergent/urgent admission, male gender, creation of a stoma, and an open approach increased the risks of all the outcomes at multivariable analysis. Age, hospital volume, hospital geographic location, and discharge to home/non-home produced different effects depending on the outcome considered. The most frequent causes of readmission were infection (19%) and bowel obstruction (14.6%).

Conclusions: We assessed national averages for mortality, LOS and readmission and related trends over a 10-year time. Laparoscopic sur-gery was the only one that could be modified by improving surgical education. Higher hospital volume was associated with a LOS reduc-tion, but our findings only partially support a policy of centralization for colorectal cancer procedures. Surgical site infection was identified as the most preventable cause of readmission.

Ó2017 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.

Keywords:Colorectal cancer; Colorectal surgery; In-hospital mortality; 30-day readmission; Length of stay

Introduction

Colorectal cancer (CRC) is the second most common malignancy in Europe with 446 800 new cases in 2012

and ranks as the second-leading causes of cancer death.1 With increases in health care costs, there is great interest in strategies that reduce costs while improving short- and long-term outcomes. Because surgery is the mainstay of CRC treatment, the evaluation of perioperative quality indi-cators is relevant, particularly if performed on large population-based databases by examining objective and comparable data.

* Corresponding author. Clinica Chirurgica I, Policlinico Universitario, via Giustiniani, 2, 35128 Padova, Italy. Fax:þ39 049651891.

E-mail address:puc@unipd.it(S. Pucciarelli). http://dx.doi.org/10.1016/j.ejso.2017.03.003

0748-7983/Ó2017 Elsevier Ltd, BASO ~ The Association for Cancer Surgery, and the European Society of Surgical Oncology. All rights reserved.

ScienceDirect

(2)

Among the many perioperative quality indicators, in-hospital mortality, unplanned readmission, and length of hospital stay (LOS) are the most studied because they are easily measured and strongly associated with both the qual-ity of care and health care spending. While the use of in-hospital mortality as a quality indicator can be misleading,2 the LOS and rates of unplanned readmission are interde-pendent outcomes that, when examined together, can help to determine the optimal path to minimizing costs and increasing health gains.3e5

Indicators of surgical quality may be used to compare performances between hospitals and may influence health care policy. For example, if optimal perioperative care is found to be associated with the surgeon or hospital volume, a policy of centralization of CRC surgery may be preferred. Moreover, because health care delivery and socio-economic background vary among countries, it is relevant to evaluate markers of quality of care in different health care systems and within each national health care context.

Despite a large number of studies that have assessed the rates of postoperative mortality and readmission and LOS after CRC surgery, the majority include only one or two of these quality indicators,3,4,6e11 have small sample sizes,8e11are not population-based,8e10and include patients with both benign and malignant colorectal diseases.3,8,12

To overcome these limitations, we conducted a national-based investigation with the aim of simultaneously evalu-ating three surgical care indicators. This study included all Italian patients with a diagnosis of primary CRC who underwent surgery from 2005 to 2014.

The principal end-points of the study were assessments of the rates of in-hospital mortality and 30-day readmission and the LOS. The predictors of each of these outcomes and the causes of 30-day readmission were also investigated.

Methods

Study design and data source

In this observational retrospective national-based study, the data were retrieved only from the administrative Na-tional Italian Hospital Discharge Dataset. The Dataset, which has been established in 1996, is utilised by the Italian Ministry of Health for administrative purpose (reimburse-ment of hospitals, based on the DRG system); the Ministry also produces a national annual report on hospital admis-sions, available on-line for epidemiologic studies and sup-plies researchers with anonymized and de-identified data from the database.14

The hospital discharge form reports of patient demo-graphics, dates of admission, surgical procedure and discharge, codes for primary and five secondary diagnoses and up to six procedures performed, surgical approach (open or laparoscopic), acuity of the admission (emergent, urgent or elective), status at discharge (died or alive), and discharge destination (home, non-home).

After approval of the study design by the Italian Minis-try of Health, we had access to the data for this specific study. The analysis and interpretation of the data are the re-sponsibility of the authors and do not represent the view of the Italian Ministry of Health.

Patients selection and definitions

Patients were identified according to the International Classification of Diseases, Ninth Revision, Clinical Modifi-cation 2007 (ICD-9-CM). The study included all patients aged 18þyears who were admitted into any Italian hospital with a diagnosis of primary colon (ICD9-CM 153.x) or rectal cancer (ICD9-CM 154.x) who underwent an elective or emergent/urgent surgical procedure from January 2005 to November 2014. The following ICD9-CM procedures were included: 45.7x, 45.8, 48.35, 48.49, 48.5, 48.6x, and 4595. The exclusion criteria were the following: discharge to acute-hospitals if the record of the second hospitalization was unavailable, cancer of the anus (ICD9-CM 154.2, 154.3), date of cancer diagnosis before January 1, 2005, and placement of a stoma before the index hospitalization. Patients who died during the index hospitalization were excluded from the analysis of readmission.

Outcome measures

The main outcomes of interest were in-hospital mortal-ity, 30-day readmission, and LOS of the index hospitaliza-tion, which was defined as the hospitalization at which the original CRC surgery was performed. In-hospital mortality was defined as the death due to any cause during the index hospitalization prior to discharge. Thirty-day readmission was defined as any unplanned, distinct hospitalization within 30 days after the discharge of the index hospitaliza-tion. This outcome has been found to be a reliable surrogate of overall unplanned readmission because most preventable readmissions occur within one month of discharge.15LOS was defined as the interval between the date of the admis-sion and the date of discharge.

The secondary outcomes were the following: the assess-ments of risk factors for mortality, readmission, and pro-longed LOS; the assessments of the cause, mortality and reoperation rate; and the LOS of 30-day readmissions.

Risk factors for in-hospital mortality, 30-day readmission and LOS

The following variables were assessed for the prediction of in-hospital mortality, 30-day readmission and LOS: age, gender, geographic area of the hospital, indexes of comor-bidity, elective or emergent/urgent admissions, site of the tumour, year of the index hospitalization, stoma creation during the index hospitalization, open or laparoscopic approach, and hospital volume of the procedures selected for this study.

(3)

Age was subdivided into five classes: 18e49, 50e59, 60e69, 70e79, and 80þyears. The geographic locations of the hospitals were divided into Northern, Central, and South/Islands areas. Comorbidities were measured as non-CRC-related hospitalizations in the year prior to the in-dex hospitalization, and admissions for abdominal non-CRC-related surgery and the Charlson Index both referred to the three years prior to the index hospitalization.16The tumour site was recorded as the colon or rectum. The me-dian annual CRC procedure volumes for each hospital and LOS were subdivided into quartiles.

Causes of the 30-day readmission

Using ICD-9CM diagnosis codes, the causes of 30-day readmission were subdivided as follows: infection, obstruc-tion, urinary tract (excluding infection), cardiovascular, haemorrhage-related (haemorrhage, anaemia and haemor-rhagic shock), malnutrition/dehydration, abdominal pain, stoma-related, neurologic, thromboembolism, other abdom-inal diagnoses, and other/unclassified. Infections were sub-divided as likely related to abdominal complications, urinary tract infection or infectious colitis, infection of the abdominal wound, sepsis, and other unspecified infections. The surgical procedures performed, modality of discharge and LOS of the readmission hospitalization were recorded.

Statistical analysis

The chi-square test was used to assess differences in de-mographics and clinical characteristics according to in-hospital mortality and 30-day readmission, and a median regression was applied for the LOS. Multilevel multivariate logistical regression was used to calculate the adjusted odd ratio (OR). The use of multilevel regression was necessary due to the hierarchical structure of the data (first level: pa-tient; second level: hospital). In the multilevel analysis, the LOS was categorized into two levels, i.e., under and over the median. The Stata 13.0 statistical package (Stata Corpo-ration, College Station, TX) was used to perform all ana-lyses. A p value of0.05 was considered significant.

Results Main outcomes

A flow-chart with the patients included in the study is re-ported inFig. 1. Among the 353 941 patients who meet the inclusion criteria, 8867 (2.5%) patients died in the hospital. Among the 345 074 patients who were discharged alive, 20 808 (6%) patients were readmitted within 30 days after discharge. The median (IQR range) LOS of the 353 941 pa-tients was 13 (9e19) days. As illustrated in Table 1 and Fig. 2, a linear decreasing was observed over time in the LOS (p¼0.0001), the rate of in-hospital mortality decreased in the last two years of observation (p¼0.016), and the rate of

30-day readmission increased from 5.8% during the 2005e06 to 6.2% during the 2013e14 (p¼0.0001).

Predictors of in-hospital mortality

Univariable analysis (Table 1) revealed that all of the variables exhibited a statistically significant association with in-hospital mortality.

On multivariable analysis (Table 2), such association was confirmed for only older age, male gender, northern/centre location of the hospital, comorbidities, stoma creation, emer-gent/urgent surgery, open approach, and colon site (Fig. 3).

Predictors of 30-day readmission

Univariable analysis (Table 1) revealed that all of the vari-ables with the exception of geographic area exhibited a statis-tically significant association with the 30-day readmission.

On multivariable analysis (Table 2), male gender, north-ern location of the hospital, comorbidities, emergent/urgent

(4)

Table 1

Main characteristics of study patients, according to each outcome evaluated, and association* with study variables.

In-hospital mortality 30-day readmission Length of hospital stay All patients Died All patients Readmitted All patients LOS

n Column % n Row % p Value n Column % n Row %

p Value n % Median IQR p Value

Patients 353 941 100.0 8867 2.5 345 074 100.0 20 808 6.0 353 941 100.0 13 (9e19) Age categories 18e49 18 209 5.1 82 0.5 0.001 18 130 5.3 1048 5.8 0.001 18 209 5.1 11 (9e16) 0.001 50e59 46 375 13.1 316 0.7 46 062 13.3 2375 5.2 46 375 13.1 11 (9e16) 60e69 94 333 26.7 1032 1.1 93 302 27.0 5011 5.4 94 333 26.7 12 (9e17) 70e79 120 946 34.2 2875 2.4 118 071 34.2 7348 6.2 120 946 34.2 13 (10e20) 80þ 74 078 20.9 4562 6.2 69 509 20.1 5026 7.2 74 078 20.9 15 (11e23) Gender Male 197 130 55.7 5160 2.6 0.001 191 971 55.6 12 634 6.6 0.001 197 130 55.7 13 (9e19) 1.000 Female 156 811 44.3 3707 2.4 153 103 44.4 8174 5.3 156 811 44.3 13 (9e19)

Geographic area of hospitals

North 181 719 51.3 4579 2.5 0.001 177 131 51.3 10 811 6.1 0.176 181 719 51.3 12 (9e18) 0.001 Centre 91 846 25.9 2496 2.7 89 355 25.9 5312 5.9 91 846 25.9 13 (9e19) South/Islands 80 376 22.7 1792 2.2 78 588 22.8 4685 6.0 80 376 22.7 15 (11e21)

Hospitalization in the year prior to the index surgery

None 246 864 69.7 5273 2.1 0.001 241 587 70.0 13 092 5.4 0.001 246 864 69.7 13 (9e19) 0.001

One 77 824 22.0 2326 3.0 75 501 21.9 5230 6.9 77 824 22.0 13 (10e19)

More than one 29 253 8.3 1268 4.3 27 986 8.1 2486 8.9 29 253 8.3 14 (10e21)

Abdominal surgery in the 3 years prior to the index surgery

No 323 377 91.4 7356 2.3 0.001 316 016 91.6 18 408 5.8 0.001 323 377 91.4 13 (9e19) 0.001 Yes 30 564 8.6 1511 4.9 29 058 8.4 2400 8.3 30 564 8.6 15 (10e22) Charlson score 0 272 277 76.9 5103 1.9 0.001 267 169 77.4 14 524 5.4 0.001 273 000 76.9 12 (9e18) 0.001 1e2 70 855 20.0 2758 3.9 68 098 19.7 5140 7.5 71 183 20.0 15 (10e22) 3þ 10 809 3.1 1006 9.3 9807 2.8 1144 11.7 10 890 3.1 17 (11e26) Admission modality Elective 261 954 74.0 3605 1.4 0.001 258 346 74.9 13 741 5.3 0.001 261 954 74.0 12 (9e16) 0.001 Emergent/urgent 91 987 26.0 5262 5.7 86 728 25.1 7067 8.1 91 987 26.0 19 (13e27)

Year of index Hospitalization

2005e2006 71 460 20.2 1833 2.6 0.016 69 625 20.2 4063 5.8 0.001 71 460 20.2 14 (10e21) 0.001 2007e2008 74 700 21.1 1887 2.5 72 812 21.1 4315 5.9 74 700 21.1 13 (10e20) 2009e2010 72 742 20.6 1824 2.5 70 916 20.6 4159 5.9 72 742 20.6 13 (9e19) 2011e2012 70 136 19.8 1816 2.6 68 323 19.8 4345 6.4 70 136 19.8 12 (9e18) 2013e2014 64 903 18.3 1507 2.3 63 398 18.4 3926 6.2 64 903 18.3 11 (9e17)

Stoma creation in the index hospitalization

No 301 777 85.3 6341 2.1 0.001 295 433 85.6 15 760 5.3 0.001 301 777 85.3 12 (9e18) 0.001 Yes 52 164 14.7 2526 4.8 49 641 14.4 5048 10.2 52 164 14.7 16 (11e25) Surgical approach Open 278 180 78.6 8384 3.0 0.001 269 793 78.2 16 935 6.3 0.001 278 180 78.6 14 (10e20) 0.001 Laparoscopy 75 761 21.4 483 0.6 75 281 21.8 3873 5.1 75 761 21.4 10 (8e14) Hospital volume 1st quartile (1e43) 92 181 26.0 2483 2.7 0.001 89 694 26.0 5545 6.2 0.002 92 181 26.0 15 (11e21) 0.001 2nd quartile (44e82) 86 009 24.3 2440 2.8 83 568 24.2 5164 6.2 86 009 24.3 13 (10e20) 3rd quartile (83e150) 87 628 24.8 2234 2.5 85 394 24.7 4976 5.8 87 628 24.8 12 (9e18) 4th quartile (151þ) 88 123 24.9 1710 1.9 86 418 25.0 5123 5.9 88 123 24.9 11 (9e17)

Length of hospital stay

1st quartile (1e9) Not applicable 89 502 25.9 3858 4.3 0.001 Not applicable

2nd quartile (10e13) 97 420 28.2 4926 5.1

3rd quartile (14e19) 76 693 22.2 4873 6.4

4th quartile (20þ) 81 459 23.6 7151 8.8

Modality of discharge

Home Not applicable 338 391 98.1 20 254 6.0 0.001 347 258 98.1 13 (9e19) 0.001

Non-home 6683 1.9 554 8.3 6683 1.9 21 (12e34)

Site

Colon 27 324 78.9 7488 2.7 0.001 271 834 78.8 15 160 5.6 0.001 279 324 78.9 13 (9e19) 0.001 Rectum 74 617 21.1 1379 1.8 73 240 21.2 5648 7.7 74 617 21.1 13 (10e20) *p Values estimated by univariable analysis.

(5)

surgery, stoma creation, open approach, rectal tumour loca-tion, and longer LOS were found to be associated with an increased risk of 30-day readmission (Fig. 4).

Predictors of LOS

Univariable analysis (Table 1) revealed that all of the variables exhibited a statistically significant association with the LOS.

According to multivariable analysis (Table 2), older age categories, male gender, southern hospital location, comorbid-ities, emergent/urgent surgery, stoma creation, open approach, rectum tumour location, non-home discharge, early period in-dex hospitalization and low hospital volume were found to be independent risk factors for a longer LOS. The ORs revealed linear correlations with the age category, the year of the index hospitalization, and the hospital volume (Fig. 5).

Overall, some variables (those indicating a high comor-bidity, emergent/urgent admission, male gender, stoma cre-ation during the index surgery, and open approach) increased the risks for all of the outcomes, and others (age, hospital volume, early period of index hospitalization, geographical locations of the hospitals, discharge at home/ non home) exhibited different effects depending on the outcome considered (Fig. 3).

Cause, reoperation and outcome of 30-day readmission

The causes, rates of reoperation and mortality, and LOS of the readmissions are summarized in Table 3. The two most frequent causes of 30-day readmission were infection

(n ¼3 962, 19%) and abdominal obstruction (n¼3 028, 14.6%). During the readmission hospitalization, 4079 (19.6%) patients underwent a surgical procedure, and 1345 (6.5%) patients died. The median (IRQ range) LOS of the readmitted patients was 8 (4e13) days.

Discussion

In-hospital mortality, 30-day readmission, and LOS have been widely used as quality indicators in colorectal surgery. The evaluation of these outcomes over a ten-year period and in a national setting was the main goal of this study.

We found an in-hospital mortality rate for CRC surgery of 2.5% (Table 1), which is comparable with the results of other similar studies in which the in-hospital or 30-day mortality rates ranged between 0.9% and 9.9%.4,6,7,10,12,13,17e22 The inclusion of patients younger than 65 and those who under-went minor colorectal procedures may explain the low rate of postoperative mortality found in the present study. A pro-gressive decrease in postoperative mortality has also been re-ported by others6and may simply reflect improvements in perioperative care over time.

We found a 30-day readmission rate of 6% that slightly increased over time (Table 1). Although the rate of readmis-sion favourable compares with the 7e25% readmission rates reported by two recent reviews and meta-ana-lyses,23,24the finding of an increasing 30-day readmission rate is unexpected. This finding may be secondary to different factors, including the selection of patients as can-didates for major surgery (presently, surgeons rarely refuse to operate on very old patients or patients with severe co-morbidities). Moreover, although early discharge has been

(6)

associated with an increased risk of the readmission among patients with multiple comorbidities or postoperative com-plications,18 in accordance with others,25 we did not observe this relationship.

The median LOS in our study was 13 days, and it decreased over time (Table 1). Because private practice is only a minor fraction of the Italian health system, our find-ings are comparable to population-based studies performed

Table 2

Multivariable analysis showing the association of study variables with in-hospital mortality, 30-day readmission and length of hospital stay. In-hospital mortality 30-day readmission Length of hospital stay

OR 95% CI OR 95% CI OR 95% CI Age categories 18e49 1.00 1.00 1.00 50e59 1.66 1.29e2.12 0.89 0.83e0.96 1.07 1.03e1.11 60e69 2.43 1.93e3.05 0.89 0.83e0.95 1.27 1.23e1.32 70e79 4.42 3.53e5.55 0.94 0.88e1.01 1.84 1.77e1.91 80þ 9.74 7.78e12.21 1.02 0.95e1.10 2.45 2.35e2.55 Gender Male 1.00 1.00 1.00 Female 0.84 0.80e0.88 0.84 0.82e0.87 0.95 0.94e0.97

Geographic area of hospitals

North 1.00 1.00 1.00

Centre 1.02 0.95e1.08 0.94 0.90e0.98 1.14 1.07e1.21

South/Islands 0.68 0.63e0.72 0.84 0.80e0.88 1.8 1.70e1.91

Hospitalization in the year prior to the index surgery

None 1.00 1.00 1.00

One 1.22 1.15e1.28 1.23 1.19e1.27 1.04 1.02e1.07

More than one 1.44 1.33e1.54 1.43 1.36e1.50 1.16 1.13e1.20

Abdominal surgery in the 3 years prior to the index surgery

No 1.00 1.00 1.00 Yes 1.14 1.06e1.22 1.08 1.02e1.13 1.07 1.03e1.10 Charlson score 0 1.00 1.00 1.00 1e2 1.43 1.36e1.51 1.22 1.17e1.26 1.34 1.31e1.36 3þ 2.84 2.61e3.08 1.69 1.58e1.81 1.79 1.70e1.88 Admission modality Elective 1.00 1.00 1.00 Emergent/urgent 3.03 2.89e3.17 1.36 1.31e1.41 4.03 3.95e4.12

Year of index Hospitalization

2005e2006 1.00 1.00 1.00

2007e2008 1.00 0.92e1.08 1.02 0.97e1.08 0.81 0.76e0.88

2009e2010 0.97 0.90e1.05 1.04 0.99e1.10 0.67 0.62e0.72

2011e2012 1.01 0.93e1.10 1.13 1.08e1.20 0.56 0.52e0.61

2013e2014 0.94 0.86e1.02 1.13 1.07e1.20 0.44 0.41e0.48

Stoma creation in the index hospitalization

No 1.00 1.00 1.00 Yes 2.37 2.25e2.50 1.68 1.62e1.74 1.9 1.86e1.95 Surgical approach Open 1.00 1.00 1.00 Laparoscopy 0.31 0.28e0.34 0.94 0.90e0.97 0.58 0.56e0.59 Hospital volume 1st quartile (1e43) 1.00 1.00 1.00 2nd quartile (44e82) 1.16 1.08e1.24 1.03 0.98e1.07 0.81 0.76e0.86 3rd quartile (83e150) 1.16 1.08e1.24 1.00 0.95e1.05 0.69 0.64e0.74 4th quartile (151þ) 0.97 0.90e1.06 1.04 0.99e1.09 0.55 0.50e0.61

Length of hospital stay

1st quartile (1e9) Not applicable 1.00 Not applicable

2nd quartile (10e13) 1.13 1.08e1.18

3rd quartile (14e19) 1.34 1.28e1.41

4th quartile (20þ) 1.68 1.60e1.76

Modality of discharge

Home Not applicable 1.00 1.00

Non-home 0.95 0.87e1.04 2.09 1.96e2.23

Site

Colon 1.00 1.00 1.00

Rectum 0.81 0.76e0.86 1.28 1.24e1.33 1.45 1.42e1.48

(7)

in countries with similar public health care.3,26Moreover, as in other reports,3,18,26 the median LOS progressively decreased over time. Improvements in postoperative hospi-tal care, a policy of cost-containment and the implementa-tion of fast-track programs27 in many hospitals likely explain this trend.

The risk for all three examined factors was independently increased according to several variables (Table 2): those re-flecting high comorbidity, emergent/urgent admission, and stoma creation. These results are in line with the findings of previous studies.3,4,7,19,21,23 Notably, the laparoscopic approach significantly reduced the risks for all the outcomes (Table 2). Although the benefits of the laparoscopic approach may be attributable to patient selection, our finding that this

approach was an independent protective factor is relevant and confirms the results of other studies.4,7,11e13,25,28e30 Consequently, efforts should be made to implement nation-wide structured training in laparoscopic surgery31 as well as the objective assessment tools for analysing technical performance.32

Other variables exhibited different associations depend-ing on the specific outcome considered. Although older age was a strong risk factor for in-hospital mortality and LOS, it did not independently affect the 30-day readmission. Indeed, although there is wide agreement that older age is a risk factor for postoperative mortality,7 its influence on 30-day readmission is controversial.3,8,23,24,26 Indeed, old patients who neither die after the surgical procedure nor

(8)

are transferred to nursing facilities or hospice tend to have characteristics similar to younger patients. Moreover, these findings may reflect the utilization of less aggressive surgi-cal approaches (i.e., losurgi-cal excision) in the elderly.

The association between hospital volume and LOS in favour of high volume hospitals has already been reported.3,17 However, like others,10,17 we did not demonstrate clear

associations of hospital volume with in-hospital mortality or 30-day readmission. Accordingly to our previous findings,17 the overall impact of this variable on short-term outcomes af-ter CRC surgery remains controversial. Further studies that include data regarding the surgical complexity of CRC pro-cedures and long-term outcomes may help to clarify this issue. Interestingly, in a recent study, a higher hospital volume

(9)

for laparoscopic surgery was found to be significantly associ-ated with a decreased 30-day re-operation rate.33Therefore, it is possible that associations between hospital volume and short-term outcomes may be evident only in subgroups of pa-tients, such as those with mid-low rectal cancer or those sub-jected to the laparoscopic approach.

In agreement with other studies,3we found that patients with colon cancer were at lower risks of 30-day readmis-sion and prolonged LOS than those with rectal cancer (Table 1). Like others,6,19,21,22,34 we also found that the location of the tumour in the colon independently increased

the risk of in-hospital mortality compared with rectal can-cer (Table 2). However, this finding requires further evalu-ation in the light of different pathologic stages of disease (this information was not available from this database) and of the inclusion of minor surgical procedures (e.g., local excisions for rectal cancer), which might actually have lowered the mortality solely in the rectal cancer group. Moreover, the anatomical definition of the rectum is known to not coincide with that of the “surgical” rectum. Surgical resections for mid-low rectal tumours are more technically demanding and pose a higher risk for post-operative

(10)

morbidity than those performed for tumours of the upper rectum or rectosigmoid junction. Notably, in line with this consideration, we found that stoma creation (usually performed in patients with mid-low rectal cancer) was a significant risk factor for all of the outcomes considered in this study (Table 2).

Although a policy of early discharge after CRC surgery raises concerns about the risk of higher rates of readmission and has been a debated topic, our findings, in agreement with others,4,23,25,35demonstrate the opposite. This finding is not surprising because a prolonged LOS is often associ-ated with postoperative complications, advanced age and preoperative comorbidity.5,18,28

We also found contradictory findings regarding the influ-ence of the geographic location of the hospital on this outcome. Although in-hospital mortality and 30-day read-mission were significantly higher in the Northern compared with the South/Islands hospitals, the opposite was found regarding LOS. Geographic variations have also been re-ported by others5 and likely depend on cultural, socio-economic and organizational factors. For example, it is customary in the southern regions to transfer patients who are dying to their homes.

In our study, approximately 20% of the readmitted pa-tients required a surgical procedure (Table 3); 6.5% of these patients died during the readmission stay, and the median

LOS was eight days. Similar to others,24,36 we found that the most frequent reasons for readmission were surgical site infections and bowel obstruction. Although one-third of readmissions have been considered as preventable,35 the data retrieved from large databases are generally not adequate to identify preventable readmissions. While read-missions of patients with a stoma and dehydration are clearly preventable and may be reduced by improving the quality of discharge instruction, the timing of the postoper-ative clinic visit, and telephone follow-up, bowel occlusion is not easily preventable. Wound infection is at least partially preventable and can be reduced through multiple interventions whose implementations should be accurately tracked.37 However, due to the lack of information about postoperative complications, we were not able to highlight any direct link between postoperative complications and preventable readmission.

Strengths and limitations

The strengths of this study are related to the large num-ber of patients included who represented all patients who underwent CRC surgery in Italy from 2005 to 2014. The data are robust and may be used as a reference for public health care organizations.

A further strength of this study is its generalizability, as it includes all the admissions that took place in a whole coun-try during a long period of time. This is slightly limited by the missed inclusion of Pull-through resection of rectum (ICD9-CM code 48.4), which has fallen into disuse in Italy but that could be still in use in other countries.

These data also provide insights into variations in CRC surgery and short-term outcomes over time and may form the basis for further studies of subgroups of CRC patients. Moreover, because a single measure of performance has been criticized as not reflecting the overall quality of CRC surgery,38 the simultaneous use of three outcome markers of CRC surgery quality adds value to our study.

The limitations of this study are related to its retrospec-tive design and, as with studies employing an administra-tive database, the lack of data that may have influenced the evaluated outcomes. The patients’ income, tumour stage, surgical complexity, postoperative complications, and surgeon volume were not available, but these parame-ters are often reported as risk factors for short-term out-comes following CRC surgery. A further limitation is related to the ICD-9 coding system, which may have gener-ated confusion in the definition of colon and rectal cancer because it does not account for the different clinical ap-proaches employed for rectosigmoid junction and high and mid-low rectal tumours. A further potential weakness may be related to the accuracy of the coding of the diagno-ses of the caudiagno-ses of readmission. However, because our findings are consistent with those of previous studies, the coding of the diagnoses of the causes of readmission was plausibly accurate.

Table 3

Main characteristics of readmissions.

N %

Readmitted 20 808 100

Causes

Infections 3962 19.0

Likely related to intra-abdominal complications

1886 e

Urinary tract or infectious colitis 519 e

Abdominal wound 215 e Sepsis 364 e Other 978 e Abdominal obstruction 3028 14.6 Urinary tract 1890 9.1 Cardiovascular 1663 8

Anemia, haemorrhage, hypovolemic shock 1286 6.2

Malnutrition, dehydration 1258 6

Abdominal pain 1019 4.9

Stoma-related 728 3.5

Neurological or psychiatric diagnoses 563 2.7

Thromboembolism 457 2.2

Abdominal diagnoses excluded infections and haemorrhage

2999 14.4

Other/unclassified 1955 9.4

Reoperation

Any reoperation 4079 19.6

Abdominal procedures 3310 15.9

Procedures on GI tract or stoma 1796 8.6 Modality of discharge In-hospital mortality 1345 6.5 Home 18 391 88.4 Non-home 1072 5.2 LOS Median (IRQ) 8 (4e13)

(11)

In conclusion, we demonstrated that the rate of in-hospital mortality for CRC procedures is low (2.5%) as is the rate of the readmission (6.5%). The LOS during the course of the study decreased from 14 to 11 days. The vari-ables associated with comorbidity increased the risks of all considered outcomes. The effects of other variables were not univocal, and in some cases, these variables exhibited opposite influences on different outcomes, which suggest that a single measure of performance may be misleading when evaluating the overall performance of CRC surgery. Our findings only partially support a policy of centralization for all CRC procedures. Finally, surgical site infection and bowel occlusion were the most frequent reasons for readmission, and the latter is partially prevent-able, which justifies financial investments in both research and performance improvement programmes.

Conflict of interest

All authors disclose any financial and personal relation-ships with people or organisation that could inappropriately influence the paper.

Role of funding source

This study was supported in part by grants from the Ital-ian Ministry of Health (Progetti ordinari di Ricerca Finaliz-zata N. RF-2011-02349645) and from the University of Padova (Progetti di Ateneo, PRAT 2015).

Acknowledgments

The authors wish to thank Dr. Flavia Carle of the Min-istry of Health e Direzione Generale della Programma-zione Sanitaria, for supplying the national dataset utilised for this study.

References

1. Ferlay J,E, Steliarova-Foucher E, Lortet-Tieulent J, et al. Cancer inci-dence and mortality patterns in Europe: estimates for 40 countries in 2012.Eur J Cancer2013;49:1374–403.

2. Pitches DW, Mohammed AM, Lilford RJ. What is the empirical evi-dence that hospitals with higher-risk adjusted mortality rates provide poorer quality care? A systematic review of the literature.BMC Health Serv Res2007;7:91.

3. Faiz O, Haji A, Burns E, Bottle A, Kennedy R, Aylin P. Hospital stay among patients undergoing major elective colorectal surgery: predict-ing prolonged stay and readmissions in NHS hospitals.Colorectal Dis 2011;13:816–22.

4. Damle RN, Cherng NB, Flahive JM, et al. Clinical and financial impact of hospital readmissions after colorectal resection: predictors, outcomes, and costs.Dis Colon Rectum2014;57:1421–9.

5. Greenblatt DY, Weber SM, O’Connor ES, LoConte NK, Liou JI, Smith MA. Readmission after colectomy for cancer predicts one-year mortality.Ann Surg2010;251:659–69.

6. Morris E, Taylor E, Thomas J, et al. Thirty-day postoperative mortality after colorectal cancer surgery in England.Gut2011;60:806–13. 7. Teloken PE, Spilsbury K, Platell C. Analysis of mortality in colorectal

surgery in the Bi-National Colorectal Cancer Audit.ANZ J Surg2016; 86:454–8.

8. Kassin MT, Owen RM, Perez SD, et al. Risk factors for 30-day hos-pital readmission among general surgery patients. J Am Coll Surg 2012;215:322–30.

9. Schrag D, Panageas KS, Riedel E, et al. Surgeon volume compared to hospital volume as a predictor of outcome following primary colon cancer resection.J Surg Oncol2003;83:68–79.

10. Borowski DW, Bradburn DM, Mills SJ, et al. Volume-outcome anal-ysis of colorectal cancer-related outcomes.Br J Surg2010;97:1416– 30.

11. Dobbins TA, Young JM, Solomon MJ. Uptake and outcomes of lapa-roscopically assisted resection for colon and rectal cancer in Australia: a population-based study.Dis Colon Rectum2014;57:415–22. 12. Zheng Z, Jemal A, Lin CC, Hu C-Y, Chang GJ. Comparative

effective-ness of laparoscopy vs open colectomy among nonmetastatic colon cancer patients: an analysis using the National Cancer Data Base. J Natl Cancer Inst2015;6:107.

13. Kang CY, Chaudhry OO, Halabi WJ, et al. Outcomes of laparoscopic colorectal surgery: data from the Nationwide Inpatient Sample 2009. Am J Surg2012;204:952–7.

14. Ministero della salute.Rapporto SDO primo semestre2015. [in Ital-ian] Available from: www.salute.gov.it/portale/temi/p2_6.jsp? id¼1237&area¼ricoveriOspedalieri&menu¼vuoto. [Last Accessed 30 November 2016].

15. Sibbritt DW. Validation of a 28 day interval between discharge and re-admission for emergency rere-admission rates.J Qual Clin Pract1995; 15:211–20.

16. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: develop-ment and validation.J Chronic Dis1987;40:373–83.

17. Burns EM, Bottle A, Almoudaris AM, et al. Hierarchical multilevel analysis of increased caseload volume and postoperative outcome after elective colorectal surgery.Br J Surg2013;100(11):1531–8. 18. Schneider EB, Hyder O, Brooke BS, et al. Patient readmission and

mortality after colorectal surgery for colon cancer: impact of length of stay relative to other clinical factors.J Am Coll Surg2012;214: 390–8.

19. Osler M, Iversen LH, Borglykke A, et al. Hospital variation in 30-day mortality after colorectal cancer surgery in Denmark: the contribution of hospital volume and patient characteristics.Ann Surg2011;253: 733–8.

20. Schrag D, Cramer LD, Bach PB, Cohen AM, Warren JL, Begg CB. Influence of hospital procedure volume on outcomes following surgery for colon cancer.JAMA2000;284:3028–35.

21. Shootman M, Lian M, Pruitt SL, et al. Hospital and geographic vari-ability in thirty-day all-cause mortality following colorectal cancer surgery.Health Serv Res2014;49:1145–64.

22. de Vries S, Jeffe DB, Davidson NO, Deshpande AD, Schootman M. Postoperative 30-day mortality in patients undergoing surgery for colorectal cancer: development of a prognostic model using adminis-trative claims data.Cancer Causes Control2014;25:1503–12. 23. Damle RN, Alavi K. Risk factors for 30-d readmission after colorectal

surgery: a systematic review.J Surg Res2016;200:200–7.

24. Li LT, Whitney LM, Donna LW, et al. Causes and prevalence of un-planned readmission after colorectal surgery: a systematic review and meta-analysis.J Am Geriatr Soc2013;61:1175–81.

25. Hendren S, Morris AM, Zhang W, Dimick J. Early discharge and hos-pital readmission after colectomy for cancer.Dis Colon Rectum2011; 54:1362–7.

26. Kelly M, Sharp L, Dwane F, Kelleher T, Comber H. Factors predicting hospital length-of-stay and readmission after colorectal resection: a population-based study of elective and emergency admissions.BMC Health Serv Res2012;12:77.

(12)

27. Gouvas N, Tan E, Windsor A, Xynos E, Tekkis PP. Fast-track vs stan-dard care in colorectal surgery: a meta-analysis update.Int J Colo-rectal Dis2009;24:1119–31.

28. Bartlett EK, Hoffman RL, Mahmoud NN, Karakousis GC, Kelz RR. Postdischarge occurrences after colorectal surgery happen early and are associated with dramatically increased rates of readmission.Dis Colon Rectum2014;57:1309–16.

29. Wang CL, Qu G, Xu HW. The short- and long-term outcomes of lapa-roscopic versus open surgery for colorectal cancer: a meta-analysis. Int J Colorectal Dis2014;29:309–20.

30. Nelson H, Sargent DJ, Wieand HS, et al. A comparison of laparoscopi-cally assisted and open colectomy for colon cancer.N Engl J Med 2004;350:2050–9.

31. Wyles SM, Miskovic D, Zhifang N, et al. Development and implemen-tation of the structured training trainer assessment report (STTAR) in the English National training programme for laparoscopic colorectal surgery.Surg Endosc2016;30:993–1003.

32. Foster JD, Miskovic D, Allison AS, et al. Application of objective clinical human reliability analysis (OCHRA) in assessment of tech-nical performance in laparoscopic rectal cancer surgery.Tech Colo-proctol2016;20:361–7.

33. Pucciarelli S, Chiappetta A, Giacomazzo G, et al. Surgical unit vol-ume and 30-day reoperation rate following primary resection for colo-rectal cancer in the Veneto Region (Italy).Tech Coloproctol2016;20: 31–40.

34. Kolfschoten NE, van Leersum NJ, Gooiker GA, et al. Successful and safe introduction of laparoscopic colorectal cancer surgery in Dutch hospitals.Ann Surg2013;257:916–21.

35. Wick EC, Shore AD, Hirose K, et al. Readmission rates and cost following colorectal surgery.Dis Colon Rectum2011;54:1475–9. 36. Merkow RP, Ju MH, Chung JW, et al. Underlying reasons associated

with hospital readmission following surgery in the United States. JAMA2015;313:483–95.

37. Cima R, Dankbar E, Lovely J, et al. Colorectal surgery surgical site infection reduction program: a national surgical quality improvement program-driven multidisciplinary single-institution experience. Am Coll Surg2013;216:24–33.

38. Almoudaris AM, Burns EM, Bottle A, et al. Single measures of per-formance do not reflect overall institutional quality in colorectal can-cer surgery.Gut2013;62:423–9.

Figure

Figure 1. Study flow-chart.
Figure 2. Crude 30-day readmission rate, in-hospital mortality rate and median length of stay, per year.
Figure 3. Risk factor for in-hospital mortality.
Figure 4. Risk factors for 30-day readmission.
+2

References

Related documents

Differences in age-adjusted and sex- adjusted 30-day and one-year all-cause mortality rates following hip fracture, as well as the length of stay of the fi rst hospital episode in

The variables assessed as potential confounders on the relationship between antithrombotic drug use and 30-day mortality risk included: age, gender, high- vs low-energy trauma,

Although truly preventable causes of 30-day readmission are low given various non-modifiable factors, this study ultimately identified common reasons for hospital visits such as

In an effort to concentrate on the PICO question, “Does a follow-up telephone call reduce the rate of 30-day hospital readmission for patients recovering from the TAVR procedure?”,

As for overall admission, the 30-day readmission data show that the percentage of elderly women who were initially admitted for a fall decreased by 2.5% between 2010 and 2014

30-day hospital readmission of older adults using care transitions after hospitalization: a pilot prospective cohort study.. Paul Y Takahashi 1 Lindsey R Haas 2 Stephanie M Quigg

The risk-adjustment models for 30-day readmission based on HDRs only and HDRs plus clinical data had a lower discrimination than the mortality models, and adding clinical data did

Impact of preoperative serum albumin on 30-day mortality following surgery for colorectal cancer: a population-based cohort study.. To view please visit the journal